Module 05 Summary

What this module established

A p-value is the share of null worlds that would produce data at least this extreme. H0 is assumed true to compute it, so it cannot be the probability that H0 is true — a simulated null gave 0.0112 where scipy gave 0.0114.

Carry forward

• Alpha delivers exactly the rate you set: 4.95% of true nulls rejected at 0.05. Across twenty tests the chance of at least one false positive is 64.5%.
• A one-sided test halves the p-value, so the direction must be fixed in advance. Chosen afterwards it makes the real error rate 10%.
• Paired versus independent can decide the result: the same 25 pairs gave p = 0.478 independent and p = 3.35e-05 paired.
• A p-value measures evidence against the null, not effect size. A difference five times smaller had by far the smaller p-value.

Before moving on

Move on when your test choice follows from four written answers and every result carries an interval.

Mark complete